Riemann Hypothesis × Claude

Claimed — independent verification in progress · updated Nov 20, 2026

Riemann Hypothesis & Claude: The Claimed Proof, Explained

On November 20, 2026, Anthropic mathematician Levent Alpöge posted a short note saying that he, working with Claude, believes he has found a proof of the Riemann Hypothesis, the 167-year-old problem about how prime numbers are spread out. He released the proof only as Lean code. No independent expert has confirmed it yet. This page tracks what was claimed, what has been checked, and how humanity got here.

The first ten non-trivial zeros of the Riemann zeta function, plotted to scale, all lying on the vertical line where the real part equals one half. 10i 20i 30i 40i 50i ½ + 14.1347… i ½ + 21.0220… i ½ + 25.0109… i 0 ½ 1 real part of s
The first ten non-trivial zeros of ζ(s), drawn to scale. The Riemann Hypothesis says every one of them, infinitely many, sits on the line Re(s) = ½.

What happened on November 20, 2026

A mathematician on Anthropic's staff posted a four-page note and a large Lean archive on his personal website. Here are the essentials, drawn from the note itself and from same-day reporting.

  1. The claim

    Alpöge wrote: "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis." He says the method also settles the Generalized Riemann Hypothesis (GRH), and expects the Extended Riemann Hypothesis to follow with more time.

  2. The evidence is code, not a paper

    Instead of a preprint, he posted a Lean formalization split into 14 archive files on his website, alpo.ge. He says he and Claude formalized the arguments without assumptions.

  3. No readable proof yet

    Alpöge says an informal draft produced with Claude ran past 250,000 pages. He chose not to post it, and plans to write the argument up properly himself.

  4. Claude did much of the heavy lifting

    He credits Claude with outpacing him near the finish line, including the idea of chaining theta lifts about √X times, a step at the heart of the argument.

  5. About six weeks

    A timeline in the note runs from the first key step on October 6, 2026 to the Generalized Riemann Hypothesis on November 16, 2026.

  6. Not yet verified

    Lean checks every logical step, but experts still have to confirm that the archive builds, that its main theorem really states RH and GRH, and that no extra axioms were added.

Verification tracker: where the claimed proof stands

A formal proof is only as good as its statement and its build. This checklist follows the steps mathematicians will expect before calling the problem solved. Last reviewed November 20, 2026.

  • Lean archive published

    Posted at alpo.ge on November 20, 2026, in 14 parts, alongside a four-page note.

    Done
  • Independent build of the Lean code

    On launch day, mathematicians posting on X reported that the archive was still compiling, a process expected to take hours.

    In progress
  • Main theorem states RH and GRH faithfully

    Reviewers must read the formal statement and every definition behind it, such as the zeta function and its non-trivial zeros.

    Pending
  • Only standard axioms, no gaps

    Lean can list every axiom a theorem depends on. That list needs to be checked and reported publicly.

    Pending
  • Human-readable write-up

    Alpöge says he intends to write the argument properly. No paper has been posted.

    Not yet
  • Expert review and broad acceptance

    For a result of this size, understanding and acceptance usually take months or longer.

    Not yet
  • Clay Millennium Prize

    Clay's rules require publication in a qualifying journal, two years, and general acceptance.

    Not yet

Levent Alpöge's account: six weeks from first idea to GRH

Levent Alpöge is a mathematician on Anthropic's staff and a co-author of the August 2026 paper on Claude's 67% result on Riemann zeta zeros. In his November 20 note he says the Riemann Hypothesis had been his private long-term goal since his student days, and that he spent years learning every tool he thought the problem might one day need.

He describes steering the work toward a setting that mirrored a simpler "toy" model, after which Claude began to outrun him. He singles out one step. To imitate the multiplicity trick in Sergei Stepanov's 1969 proof of the Riemann Hypothesis over finite fields, Claude proposed chaining theta lifts, and suggested a chain whose length grows like √X for a prime sum of length X.

The note ends with a 20-step timeline that, Alpöge writes, Claude assembled after searching their work. A few of the steps:

  • October 6, 2026: a "Stepanov determinant", the first key idea.
  • October 20: the argument commits to a weighted sum over primes built from the von Mangoldt function.
  • November 15: iterated theta lifts between unitary groups supply the multiplicity the argument needs.
  • November 16: the Generalized Riemann Hypothesis.

Alpöge is open about the caveats. He calls the argument extremely long, says he may have baked in strange, suboptimal choices, and expects much shorter proofs to follow. He also writes that he has not yet mastered all of the mathematics Claude brought in, especially automorphic forms on higher-rank groups.

The Riemann Hypothesis in one minute

Prime numbers look scattered at random, yet their overall count follows a smooth curve. In 1859 Bernhard Riemann showed that the exact error in that curve is controlled by the zeros of a single function, the zeta function ζ(s). He guessed that every "non-trivial" zero lies on one vertical line in the complex plane, where the real part equals ½. If he was right, the primes are as evenly spread as they can possibly be.

Computers have confirmed that more than 12 trillion zeros sit on the line, but no amount of checking can prove a statement about infinitely many zeros. Our plain-English guide to the Riemann Hypothesis walks through it step by step.

How mathematicians reacted on day one

Early reactions mixed excitement with caution. According to a roundup by OfficeChai, mathematician Elliot Glazer said the formal statement in the release is GRH and that the code was still compiling. Thomas Bloom of the University of Manchester pointed people to the Lean community's Zulip chat, where experts were examining it. Ayush Khaitan of Princeton argued that Alpöge deserves as much credit as Claude, and mathematician and author Kit Yates stressed that a claimed proof is not yet an accepted one.

Scientific American noted that problems can arise in Lean, but said experts should not need long to check whether the hypothesis is coded correctly and whether the proof holds.

What happens next

The next milestones are concrete: an independent build of the Lean archive, a check that its main theorem really states RH and GRH with standard definitions, a public list of the axioms it uses, and eventually a human-readable account. We explain each step in Has the Riemann Hypothesis been solved? and update the tracker above as verifiable news comes in. For the long view, from Euler to Claude, read the Riemann Hypothesis timeline.

If the proof holds up, what changes?

The Riemann Hypothesis sits underneath a large part of modern number theory. If the Lean proof is confirmed, these are the most direct consequences.

The primes become as regular as possible

In 1901 Helge von Koch showed that RH is equivalent to the best possible error term in the prime number theorem: the prime count stays within about √x · log x of Gauss's estimate li(x).

Conditional theorems become theorems

For over a century, number theorists have proved results that begin "assuming RH" or "assuming GRH". A verified proof would make all of them unconditional at once.

GRH answers questions in computing

Under GRH, Gary Miller's 1976 primality test is deterministic and fast, and the least quadratic non-residue modulo a prime is provably small. (The AKS test of 2002 already gives fast deterministic primality testing without GRH.)

It does not break encryption

RSA and similar systems rely on how hard it is to factor large numbers. A proof of RH gives no new way to factor, so online banking is unaffected.

From Riemann's 1859 paper to Claude

  1. Riemann conjectures that every non-trivial zero of ζ(s) lies on the line Re(s) = ½.

  2. Hilbert includes it in his famous list of problems for the new century.

  3. Hardy proves that infinitely many zeros lie on the line.

  4. Levinson shows at least a third do; Deligne proves the finite-field analogue.

  5. The Clay Mathematics Institute offers $1 million for a proof.

  6. Claude lifts the proven share of zeros on the line to over two-thirds.

  7. Alpöge announces a claimed proof of RH and GRH, made with Claude.

Explore the full Riemann Hypothesis timeline

Frequently asked questions

Has Claude solved the Riemann Hypothesis?

Not confirmed. On November 20, 2026, Levent Alpöge announced that he and Claude had found a proof, and released it as Lean code. Until independent experts confirm that the code builds and proves the right statement, it remains a claimed proof.

Who is Levent Alpöge?

A mathematician employed by Anthropic. With Ralph Furman he co-authored the August 2026 paper presenting Claude's result that more than two-thirds of the zeta zeros lie on the critical line. Scientific American reports that he shared a Claude-generated result on the Jacobian conjecture in July 2026.

What exactly did Claude do?

According to Alpöge's note, he steered the work toward the right setting and Claude supplied many of the technical steps, outpacing him near the end. He says the two of them also produced the Lean formalization together. The announcement came from Alpöge personally, on his own website.

What is Lean, and why does it matter here?

Lean is a proof assistant: software that checks every logical step of a proof against a small trusted core. If the archive builds, the logic is sound, but people still have to confirm that the theorem it proves really is the Riemann Hypothesis.

Does the claim cover the Generalized Riemann Hypothesis too?

Yes. Alpöge says the method resolves GRH, which contains the original hypothesis as its simplest case, and he expects the Extended Riemann Hypothesis to follow with more time.

Will anyone receive the $1 million Clay prize?

Not soon. The Clay Mathematics Institute considers a solution only after it appears in a qualifying peer-reviewed outlet, at least two years pass, and it gains general acceptance in the mathematics community.

Is this the same as Claude's 67% result from August?

No. In August 2026 Claude proved that more than two-thirds of the zeros lie on the critical line, a partial result that experts have checked. The November claim is a full proof, reached by a different route.

Why does the Riemann Hypothesis matter?

It pins down how evenly prime numbers are spread out. A large body of results in number theory assume it, so a proof would settle all of them at once.

Sources

  1. Levent Alpöge, statement on the Riemann Hypothesis (PDF) · Nov 20, 2026
  2. Scientific American: Anthropic mathematician claims to prove Riemann hypothesis · Nov 20, 2026
  3. OfficeChai: How mathematicians have reacted to the claimed proof · Nov 20, 2026
  4. Anthropic: Learning more about Claude's mathematical capabilities · Aug 10, 2026
  5. Alpöge & Furman, More than two thirds of the zeta zeros are simple and on the critical line (arXiv:2608.13637) · Aug 2026
  6. Clay Mathematics Institute: Millennium Prize rules

Discuss the claimed proof

What do you make of it? Questions, context and corrections are welcome. Links are not accepted, so name your source (publication and date) in plain text.